© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 34-GEMMASCOPE-RES-131K
    7. 63981
    Prev
    Next
    INDEX
    Explanations

    focuses on emphasis

    np_acts-logits-general · gemini-2.5-flash-lite

    The neuron is keyed to clause-boundary punctuation and markers—especially commas, closing parentheses, and relative pronouns like “which” that signal the start of subordinate clauses.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     terpaksa
    -1.04
    När
    -1.00
    entir
    -0.96
    Only
    -0.95
    كات
    -0.93
    isteren
    -0.91
     hubieran
    -0.90
    zaamheid
    -0.89
    ΤΟ
    -0.88
    Dette
    -0.88
    POSITIVE LOGITS
     emphasis
    2.53
     focus
    2.48
    emphas
    2.30
     focuses
    2.27
     emphas
    2.13
    focus
    1.84
    emphasis
    1.78
     advocate
    1.76
     emphasize
    1.74
     Emphasis
    1.73
    Activations Density 0.114%

    No Known Activations